Solving the Heat Transfer Mystery: My Numbers Not Equally Out?

[fiberguy02]

Can anybody explain why my numbers are not equally out?

At work, I am trying to determine the heat transfer taking place in a water cabinet. Both loops have water in them. There is a counter current flow heat exchanger where the transfer takes place.

I take the temperature measurements with a Microscanner D1001 IR thermometer. With this device I contact the surface of a pipe and take the measurement. I take about 3-5 measurements and average them out. I do this at the same spot on the pipes for the other inlet/outlet to the heat exchanger. Each pipe has the same diameter and the same material. (I know this is not a 100% correct method, but I am hoping it gets me close enough, because the thermocouple connected to the pipes are not calibrated. I have done this step about 50 times to get enough values to minimize any outliers.

Next I take the flow on the two loops.

From there I use the equation:
Heat tranferred = specific heat of water * flow * delta temperature. (I have no issues with the units).

When I compare the two loops, the numbers are off on average of 5kW (the average of the loops is (30kW and 25kW). The same loop is almost always higher. And this is the same case when I take the measurements off of duplicate machines. I got my data from over 5 water cabinets that were the same.

Why are my two loops not equally? Any guesses?

Thanks!

 

[russ_watters]

Well, how exactly are you measuring the temps and flows? How are you attaching the thermocouple? Is the piping insulated? Where are the flow numbers coming from? How far from ambient are the temperatures?

 

[fiberguy02]

I am measuring the flow with a paddle wheel flow meter.

I am measuring the temperature with a Microscanner D1001 IR thermometer.

The thermocouple is not accurate so I am not using it.

The piping may/may not be insulated. Regardless, it is the same piping for both loops inlet and outlet, so the two heat transfer equations should equal.

the temperatures measured from the pipe are between 70-94 Farenheit. Room temperature is about 75-80 F. However, I am placing the IR microscanner directly on the pipe so that the surrounding air should not play a part.

 

[Q_Goest]

the temperatures measured from the pipe are between 70-94 Farenheit. Room temperature is about 75-80 F. However, I am placing the IR microscanner directly on the pipe so that the surrounding air should not play a part.

There is a temperature gradient between the liquid and surface of the pipe because of conduction across the pipe wall. It isn't huge, but it could easily make up 5 kW out of 25 to 30 kW. Wrap an inch or more of insulation around the pipe and use a thermocouple to help eliminate this error.

There could be other errors in the way the measurements are being done, so if you can better describe what you're doing (pictures go a long way too), you'll get better responces. For example, stainless steel or reflective surfaces produce substanial error in IR thermometer. I'd suggest using a thermocouple, thermistor, or even a mechanical thermometer of some sort. I've not had tremendous luck with those infrared things.

 

[FredGarvin]

I second the IR thermometer woes. I have used them successfully ONLY when I have had a very good knowledge of the emissivity of the surface we were measuring. In your case if there is no real calibration or correction to your actual conditions, I would say that the majority of your error is in using the IR method. I would also recommend using a proper skin temp TC.

 

[fiberguy02]

But, if i take the same measurement (at the same place every time on the same surface) with the IR thermometer, shouldn't all/most of the error be canceled out when i compare one loop to the other?

 

[stewartcs]

I take the temperature measurements with a Microscanner D1001 IR thermometer. With this device I contact the surface of a pipe and take the measurement. I take about 3-5 measurements and average them out. I do this at the same spot on the pipes for the other inlet/outlet to the heat exchanger. Each pipe has the same diameter and the same material. (I know this is not a 100% correct method, but I am hoping it gets me close enough, because the thermocouple connected to the pipes are not calibrated. I have done this step about 50 times to get enough values to minimize any outliers.

I don't believe you are minimizing the outliers. You're only taking 3-5 observations for 50 random samples. In order to have a normal distribution for a random sample you need typically 30 observations, not 3-5. In other words you'll need 50 random samples of at least 30 observations (with 30 your guaranteed normality by the CLT) to have a normal sampling distribution. Then you can better assess the outliers.

Hope this helps.

CS

 

[russ_watters]

But, if i take the same measurement (at the same place every time on the same surface) with the IR thermometer, shouldn't all/most of the error be canceled out when i compare one loop to the other?

No. The temperature gradient is significant and it depends on the difference between the fluid and ambient temperatures. Ie, the 70F fluid will show a near 70F pipe surface temp. The 95F fluid might only show a 90F pipe surface temp.

I've run into this exact issue before - I only use an infrared thermometer for a spot check. When accuracy is important, I use a thermister and I always insulate the pipe for a few inches above and below the sensor.

 

[fiberguy02]

russ, that makes sense. i had never heard of a thermistor before, but i am currently looking at it now.

thanks.

 

[fiberguy02]

i think i have a plan. anybody see an issue with it?

i am going to bring the machine down, and close the valve allowing water to it.

there are already thermocouples hooked up to the system, but for whatever reason the temperatures displayed on the gauges are not accurate. if i hook up a fluke meter to the thermocouple, this should provide an accurate reading, correct?

for the flow, i am going to hook up a circuit setter to get a more accurate flow.

i will be collecting the trend for both temperature flow.

any issues?

 

[russ_watters]

You mean a circuit setter with pressure ports for a hydronic manometer, right? That should give you an accurate flow reading, yes. For the existing thermocouples, I don't know that I'd trust them - if they give you bad readings now, it could be something with the installation.